Scene Classification

Our scene classification and analysis API identifies the apparent location of the scene in a given image or video. The system returns multiple scene labels and classifies them at scale. The model checks whether it is an indoor or outdoor space and groups the scene under possible categories and attributes, such as, outdoors on a farm, around sunny mountains, etc.

The scene analysis technology can be used in cases when it’s necessary to recognize the environment at one glance. For example, people who work with many images can use the system to quickly classify the pictures according to the content; providers of stock photos and videos can apply this technology to automatically generate labels for their content.

Try Model

Main Function

Scene classification API is used to detect indoor and outdoor scenes in an image. The scene detection model gives the 5 highest possible locations out of 365 classes and provides supporting information from a dataset of 102 attributes in total.

Input

The input is an image.

Output

The API returns the following: 

Indoor/Outdoor info

  • True – Indoor
  • False – Outdoor 

Categories – Gives the 5 highest possible locations. There are a total of 365 possible locations.

Example:

  • Wheat Field 
  • Hay Field
  • Corn Field
  • Farm 
  • Pasture

Attributes- Gives additional information about a location, such as warm, natural light, etc. There are a total of 102 attributes.

Example:

  • Open area
  • Natural
  • Natural Light
  • Vegetation
  • Far-away Horizon
  • Grass
  • Camping 
  • Sunny
  • Foliage

For example

{
'Indoor': False, 
'categories': [
	'alley',
	'street', 
	'downtown',
	'arcade',
	'crosswalk'
 	 ], 
'attributes': [
	'man-made', 
  	'no horizon',
  	'natural light', 
	'vertical components', 
 	'driving', 
  	'pavement',
	'asphalt',
  	'open area',
  	'biking'
       ]
}
Go to Top